Morgan Stanley Maps a $1.58 Trillion AI Spending Wave: About a Quarter of 2027 S&P 500 Sales Growth, 78% of Earnings Growth and a Rising Debt Bill

Morgan Stanley has put a number on how much of Wall Street’s profit outlook now rests on one theme. In a new forecast, the bank expects global AI infrastructure spending by hyperscalers and neocloud providers to rise from about $980 billion in 2026 to $1.58 trillion in 2027, a jump of roughly 60%, and to reach $1.81 trillion in 2028. That spending is projected to account for about 5% of S&P 500 sales next year and roughly a quarter of the index’s revenue growth, while AI enablers and adopters are expected to drive about 78% of S&P 500 earnings growth in 2027. The flip side is the bill: roughly half of the spending is expected to be financed with debt, and leases are hiding part of it.

The numbers behind the forecast

The jump from $980 billion to $1.58 trillion is larger than it looks because Morgan Stanley now counts leases. Adding lease obligations that were previously left out raised its 2027 total by about 14%. The pace also slows: growth of roughly 60% in 2027 shrinks to about 15% in 2028 by our math, and the bank expects capex growth to decelerate after 2027.

Where the money lands

The report estimates that roughly $1 trillion of the 2027 spending, about 65% of the total, will show up as revenue at S&P 500 companies, and that Nvidia (NVDA) alone would capture about $690 billion of it, around two-thirds of that flow. Semiconductors and servers account for about 42% of 2027 AI capex. The main spenders are the hyperscalers: Microsoft, Alphabet, Amazon, Meta and Oracle. For comparison, Nvidia’s own CFO said on the company’s most recent earnings call that she expects hyperscaler capex to approach $800 billion in 2026 and reach about $1.3 trillion in 2027, with a cloud backlog above $2 trillion. Morgan Stanley’s total is broader because it also includes neocloud providers and leases.

Already visible in third-quarter earnings

The earnings effect is already large. In Morgan Stanley’s chart, AI beneficiaries and the Magnificent Seven account for 61.4% of projected third-quarter S&P 500 earnings growth, with Micron alone contributing 18.8% and Nvidia 15%. That chart assumes about 27% earnings growth, while FactSet’s latest blended estimate as of Oct. 16 is 29.0%, slightly down from 29.4% a week earlier, on revenue growth of 12.6%. So far only 9% of S&P 500 companies have reported, and 84% of them have beaten earnings estimates, above the five-year average of 78%. FactSet puts calendar 2026 earnings growth at 32.8% and the forward price-to-earnings ratio at 19.0, below its five-year average of 19.8. Goldman Sachs has estimated that AI-related capex has driven nearly half of recent S&P 500 earnings growth.

The catch: leases and debt

Morgan Stanley’s forecast includes $195 billion of lease capex in 2027, rising to $246 billion in 2028. Oracle leads the lease group at about $57.2 billion, followed by Microsoft, Meta, Amazon and Google, and hyperscalers hold more than $1.1 trillion of pending lease commitments. Leases work like debt-financed purchases but are not recorded as capex, so cash-flow statements can understate the buildout. Roughly half of all projected AI investment is expected to be debt-financed, and spending is front-loaded, which squeezes free cash flow even though the hyperscalers’ strong credit ratings give them room to borrow.

The weaker links are the heavily indebted developers. This week CoreWeave’s bond yields spiked and its shares fell 13% on Oct. 14, dragging other neocloud and AI infrastructure names lower, and a hot inflation report pushed the 10-year Treasury yield to about 5.28%. Rising borrowing costs and credit-market stress could threaten expansion plans, particularly among those providers. The same pressure showed up in the news this weekend, when Microsoft was reported to have canceled a lease at an Nscale site that could not get power.

What it means for investors

Two things can be true at once. The spending wave is real, it is already showing up in revenue and profits, and the biggest suppliers sit inside the S&P 500. But the index is also leaning on a single theme: with AI enablers and adopters supplying about 78% of 2027 earnings growth, any slowdown in spending, any financing hiccup or any disappointment in what customers earn on the spending would hit the index’s growth rate directly. Excluding AI spending, Morgan Stanley’s 2027 revenue growth estimate falls from 8.6% to 6.6%, a two-point gap that shows how much depends on it.